Permanent El Niño-Like Conditions During the Pliocene Warm Period

See allHide authors and affiliations

Science  29 Jul 2005:
Vol. 309, Issue 5735, pp. 758-761
DOI: 10.1126/science.1112596

This article has a correction. Please see:


During the warm early Pliocene (∼4.5 to 3.0 million years ago), the most recent interval with a climate warmer than today, the eastern Pacific thermocline was deep and the average west-to-east sea surface temperature difference across the equatorial Pacific was only 1.5 ± 0.9°C, much like it is during a modern El Niño event. Thus, the modern strong sea surface temperature gradient across the equatorial Pacific is not a stable and permanent feature. Sustained El Niño-like conditions, including relatively weak zonal atmospheric (Walker) circulation, could be a consequence of, and play an important role in determining, global warmth.

The low-latitude Pacific Ocean provides a substantial portion of the global atmosphere's sensible and latent heat and is thus a central driver of climate (1). Over the past 25 years, the mean equatorial Pacific sea surface temperature (SST) has increased by ∼0.8°C (2), possibly in response to increasing greenhouse gas concentration (3). Changes in the tropical Pacific mean climatic state may influence the amplitude of interannual, or El Niño-Southern Oscillation (ENSO), climate variability (4-7), which may in turn play a role in global warming (8). The tropical Pacific mean state can be influenced by extratropical conditions, where surface water is subducted and flows into the tropical thermocline (the steep subsurface vertical thermal gradient between warm surface and cooler deep waters) (4, 7, 9-12). Conversely, the mean state of the tropical Pacific could be determined by changes in ENSO variability itself (13, 14). Overall, the mechanisms that control the mean state of the tropical Pacific are not fully understood, and predictions of future change in the mean state do not agree, probably because ENSO dynamics are not well-represented by most general circulation models (15). Thus, observational studies are needed to add additional constraints on the interplay between mean tropical conditions and global climate change.

Because instrumental (directly measured) records of climate change are relatively short (∼100 years), geological records must be used to test theories that link long-term global climate change with tropical conditions (7). Characterizing conditions during times of global warmth requires investigation of older geologic periods that were substantially warmer than today. The most recent such interval is the Pliocene warm period [∼4.5 to 3.0 million years ago (Ma)], which was characterized by ∼3°C higher global surface temperatures relative to today (16). Although not a direct analogy of future transient global warming, the Pliocene warm period is a relevant natural experiment that can be used to understand processes contributing to long-term global warmth, because many boundary conditions were similar to today, including first-order ocean circulation patterns, the Earth's continental configuration, small Northern Hemisphere ice coverage, and atmospheric carbon dioxide concentrations (about 30% higher than pre-anthropogenic values) (16).

The equatorial west-to-east SST gradient and thermocline depth are intimately coupled parameters that exert a significant influence over both the mean state and variability of tropical Pacific climate (4, 7). Today, tropical trade winds drive ocean currents westward, resulting in a thick, warm, mixed layer and deep thermocline in the western equatorial Pacific (WEP) and a thin, warm, mixed layer and shallow thermocline in eastern equatorial Pacific (EEP). The trades also drive surface water divergence and upwelling of warm water in the WEP and cold water in the EEP, where the thermocline is deep and shallow, respectively. Thus, the trade winds cause a west-to-east, or zonal, asymmetry in thermocline depth, SST (Fig. 1), and surface air pressure, which in turn strengthens the winds and further augments this asymmetry. The atmospheric circulation (Walker) cell, including easterly trade winds, rising air in the west, westerly winds aloft, and sinking air in the east, is a persistent feature of the tropical Pacific today; the magnitude of the zonal SST gradient is an excellent diagnostic of the strength of Walker circulation. Extreme temporary reductions in the zonal SST gradient and Walker circulation, or El Niño events, occur every 2 to 7 years and dramatically influence global climate by redistributing heat stored in the tropical Pacific to extratropical latitudes (7, 15). Likewise, changes in the long-term SST gradient may have altered extratropical conditions for sustained periods of time in the past (17) and could also potentially influence global climate in the future (9).

Fig. 1.

Sites used in this study, ODP site 847 (0°N, 95°W, 3373-m water depth) and ODP site 806 (0°N, 159°E, 2520-m water depth), overlaid on a map of climatological mean SSTs in the tropical Pacific Ocean (36, 37).

To represent the east-west Pacific SST gradient, we used the difference in SST and δ18O between Ocean Drilling Program (ODP) site 806 in the WEP and ODP site 847 in the EEP (Fig. 1). Our sites are ideally located to monitor changes in equatorial upwelling and thermocline depth (away from the confounding effects of the Peru-Chile upwelling system). At each site, we made paired δ18O and magnesium-to-calcium ratio (Mg/Ca) measurements on foraminiferal shells (a surface-dwelling species) to construct a time series of δ18O and SST [using the calibration of Dekens et al. (18)], with an average sampling interval of 10 kyear from 5.3 Ma to present (19). The SST and δ18O records, smoothed to remove glacial-interglacial variability, indicate that from 5.3 to 1.7 Ma the west-to-east difference in both parameters (Fig. 2, A and B) was relatively small: the west was ∼2°C colder and the east ∼2°C warmer than modern SSTs, and the west-to-east SST difference was always less than 2°C, with an average of 1.5 ± 0.9 (Fig. 2C). After 2.5 Ma, SST in the EEP began gradually decreasing. At 1.7 Ma, WEP temperatures warmed by ∼2°C over a 50-kyear period, while the EEP continued to cool. By 1.6 Ma, the modern zonal SST difference of ∼4°C, equivalent to a Mg/Ca contrast of 35%, was established. Thereafter, the EEP gradually cooled, and the mean west-to-east SST difference for the last 1.6 myear was 5.1 ± 0.9°C.

Fig. 2.

Equatorial Pacific isotopic and temperature records. (A) oxygen isotope records of G. sacculifer (without sac, 355 to 425 μm) at ODP site 847 (blue) and 806 (red) as well as of G. tumida (355 to 425 μm) at ODP site 847 (green). (B) SSTs estimated from Mg/Ca measurements in G. sacculifer (without sac) from sites 847 (blue) and 806 (red). (C) The estimated zonal SST gradient on the equator between 159°E and 95°W. (D) The difference (Δ°C) between the calcification temperatures of G. sacculifer (without sac) and G. tumida at site 847 calculated by assuming that the difference between their δ18O entirely reflects temperature and that Δδ18O/0.21 = Δ°C. Heavy lines in all figures represent a 0.2-Ma Gaussian weighted running mean. Curves in (C) and (D) were calculated by using the smooth curves in (A) and (B).

Our results contradict a recent study (20), using the same methodology and site selection (19), that concluded that the EEP was cooler (not warmer) in the early Pliocene. In the Rickaby and Halloran study (20), SST changes over the last 5 myear were represented by six data points, with only one data point from the Pliocene warm period interval, and we suspect that aliasing of higher frequency (orbital scale) variability led those authors (20) to erroneously conclude that the EEP was colder in the Pliocene warm period than at present. Our higher temporal resolution study, using over 400 data points over the last 5 myear, provides a more accurate reconstruction of mean SST trends and firm evidence for warm SSTs in the EEP during the Pliocene warm period. Furthermore, warmer-than-present alkenone-based SST estimates from the EEP (21) and foraminiferal δ18O records (Fig. 2A) (16, 22, 23) indicating reduced hydrographic differences across the Pacific during the Pliocene warm period are consistent with the interpretations of our Mg/Ca records (Fig. 2C).

To track changes in the mean thermocline depth at the EEP site, we used Δδ18O, the difference in δ18O between surface-dwelling Globoritalia sacculifer (without sac) and G. tumida (355 to 425 μm) (Fig. 2A), which occupies the base of the photic zone (at ∼100 m depth) (24, 25). Because there is a weak vertical salinity gradient in the EEP today, we assume that Δδ18O, on first order, represents the difference in temperature [ΔT between the surface and ∼100m (Fig. 2D)]. High ΔT indicates that the cool thermocline water was shallow and above the base of the photic zone, whereas low ΔT indicates that cool thermocline water was deep and below the base of the photic zone. The increase in Δδ18O of ∼1.2 per mil (‰) (Fig. 2A) equates to an increase in ΔT of ∼5°C (26) (Fig. 2D), indicating the presence of a warmer or deeper thermocline in the beginning of the Pliocene, significant shoaling or cooling of the thermocline from 5.3 to 3.5 Ma, and relatively constant conditions from 3.5 Ma to present.

Previous work, which supports our thermocline reconstructions, indicates that δ18O differences between depth-stratified foraminiferal species (22, 27), foraminiferal species assemblages (27), which are strongly correlated to thermocline depth (25), and Mg/Ca-derived subsurface temperatures (20) were similar on opposite sides of the basin during the warm Pliocene and became dissimilar by ∼3.5 Ma. Taken together with our Mg/Ca-based evidence for a reduced west-to-east SST difference, these data indicate that the Pliocene warm period was not characterized by the typical west-to-east asymmetric conditions of the modern equatorial Pacific (Fig. 1). Rather, the Pliocene warm period had permanent El Niño-like conditions in several important aspects: Relative to today, the equatorial upwelling region of the EEP was warmer (Fig. 2B), the west-east SST difference along the equator was reduced (Fig. 2C), the thermocline in the EEP was deeper (Fig. 2D), and subsurface conditions were more symmetric across the tropical Pacific (20, 27). These observations are consistent with each other: A warmer and/or deeper thermocline would have resulted in warmer SSTs in EEP upwelling regions and reduced zonal SST and surface air pressure gradients. The reduced pressure gradient would have caused the winds and Walker circulation to slacken, thereby reinforcing the effects of warmer thermocline waters (28). Weak Walker circulation would have influenced the position and intensity of extratropical high- and low-pressure centers (29) and therefore would have had far-reaching climatic effects. In fact, the global expression of Pliocene warmth resembles the teleconnection pattern of El Niño (30).

The observed El Niño-like mean state during the Pliocene warm period could be related to changes in the mean state of extratropical regions (7). Theoretically, reduced subtropical SST (11) or surface salinity (12) gradients could have resulted in a warmer and/or deeper tropical thermocline and, consequently, warm water upwelling in the EEP. Although there is observational evidence for warmer subtropical SSTs (31), more detailed information on subtropical SST and salinity gradients is needed. Alternatively, the mean state could have been influenced by processes within the tropics, such as changes in the character of short-term (ENSO) variability (13, 14) perhaps due to the slightly different global boundary conditions, such as atmospheric CO2 concentrations, of the early Pliocene compared to today.

Our sites alone cannot be used to determine whether the strong north-south SST gradients (Fig. 1) changed over time; future work detailing the spatial patterns of the tropical Pacific conditions could be used to test theories of what may have caused observed changes in equatorial conditions. In addition, future studies that try to differentiate between extratropical and within-tropical impacts on the mean state of the tropical Pacific should consider that several eastern boundary regions (California, Peru-Chile, West African margins), where cool upwelling occurs today, were significantly warmer before ∼ 3 Ma (21, 32), suggesting that the thermocline was deeper and/or warmer globally, and not just in the tropical Pacific.

The difference in timing between the decrease in thermocline depth before ∼4.0 Ma (Fig. 2D) and the increase in zonal SST difference at ∼1.7 Ma (Fig. 2C) could be explained by several factors. First, because the thermocline depth proxy, Δδ18O, depends on the δ18O of G. tumida, which has a depth ecology at the base of the photic zone at ∼100 m, it may not have been sensitive to additional shoaling above 100 m, which could have occurred after 4.0 Ma. Second, changes in salinity (and associated δ18O) of subsurface water could have a secondary effect on Δδ18O; thus, the effect of an increase in ΔT on Δδ18O after 4.0 Ma could be masked by a synchronous decrease in salinity of subsurface water. Detailed records of paired Mg/Ca and δ18O measurements on G. tumida are needed to better assess ΔT (33). And third, thermocline depth and SST are not linearly related (4), because the air-sea feedbacks that cause changes in Walker circulation become stronger as the thermocline shoals. Thus, the increase in the zonal SST difference after ∼1.7 Ma could indicate that tropical air-sea feedbacks (34) amplified the SST response to a changing thermocline once critical thermocline conditions were reached.

Mean tropical thermocline conditions can influence air-sea feedbacks that affect high-frequency climate variability (4, 15); the amplitude of ENSO variability is dampened when the thermocline is deeper or warmer in the EEP (5, 10). This effect applied to longer time scales may explain why permanent El Niño conditions during the Pliocene were accompanied by reduced-amplitude glacial-interglacial cycles; a deeper or warmer thermocline may have preconditioned the tropical system such that air-sea feedbacks needed to amplify small perturbations in solar forcing were weak. The establishment of Walker circulation at ∼1.7 Ma coincides with the Pliocene-Pleistocene epoch boundary, after which the sensitivity of climate to solar forcing peaked (16).

Our study indicates that today's zonally asymmetric SST pattern and thermocline structure of the tropical Pacific are not stable over long time scales. Given the importance of tropical Pacific processes in modulating meriodional heat transport, these results indicate that in a warmer world, the ocean may accomplish redistribution of heat in a fundamentally different way. Thus, the Pliocene warm period provides a target and a test to climate models and theory and is an indication that climate feedbacks do not work to maintain the presently strong asymmetry across the Pacific under some circumstances. It may indicate that warming cannot continue indefinitely without substantial changes in the Walker circulation (10) and that changes in the subtropics, communicated through the thermocline, might cause a fundamental reorganization of the tropical Pacific ocean-atmosphere system (4, 10). Depending on one's interpretation of the instrumental data from the tropical Pacific, a shift in the baseline tropical Pacific pattern may already be occurring (2, 4, 5, 8).

Supporting Online Material

SOM Text

Materials and Methods

Tables S1 to S3

References and Notes

References and Notes

View Abstract

Navigate This Article